* data/bison.m4 (b4_percent_define_check_kind): New.
Use it to check api.token.prefix.
* data/c++.m4: Check the kind of api.namespace.
* doc/bison.texi: Update a reference to former 'namespace' variable.
* tests/input.at ("%define" code variables): Check api.namespace.
Suggested by Joel E. Denny.
http://lists.gnu.org/archive/html/bison-patches/2013-03/msg00016.html
* data/bison.m4 (b4_percent_define_get_kind): New.
(b4_variant_flag): Check that api.value.type is defined as the 'variant'
keyword value.
* data/c.m4 (_b4_value_type_setup_keyword): New.
(b4_value_type_setup): Use it to simplify reading.
Use b4_define_silent.
Decode api.value.type, including its type.
(b4_value_type_define): Likewise.
* data/c++.m4 (b4_value_type_declare): Adjust the decoding of api.value.type,
taking its kind into account.
* doc/bison.texi: Adjust all the examples to the new syntax.
* NEWS: Ditto.
* tests/types.at: Adjust
* data/c.m4 (b4_symbol_type_register, b4_type_define_tag)
(b4_symbol_value_union, b4_value_type_setup_union)
(b4_value_type_setup_variant, b4_value_type_setup):
New.
(b4_value_type_define): Use it to set up properly the type.
Handle the various possible values of api.value.type.
* data/c++.m4 (b4_value_type_declare): Likewise.
* data/lalr1.cc (b4_value_type_setup_variant): Redefine.
* tests/types.at: New.
Exercise all the C/C++ skeletons with different types of
api.value.type values.
* tests/local.mk, tests/testsuite.at: Use it.
* doc/bison.texi (%define Summary): Document api.value.type.
* NEWS: Advertise it, together with api.token.constructor.
This is to match the names used in C and api.value.type, even if the
parser actually defines semantic_type.
* data/c++.m4 (b4_semantic_type_declare): Rename as...
(b4_value_type_declare): this.
* data/variant.hh: Likewise.
"stype" is quite unclear, and it also collides with the former %define
variable that had the same name (replaced by api.value.type).
* src/parse-gram.y (stype): Rename as...
(union_members): this.
* data/bison.m4: Adjust.
(b4_user_stype): Rename as...
(b4_user_union_members): this.
* data/c++.m4, data/c.m4: Adjust.
* src/parse-gram.c: regen.
Recently, there was a slightly vicious bug hidden in the make_ functions:
parser::symbol_type
parser::make_TEXT (const ::std::string& v)
{
return symbol_type (token::TOK_TEXT, v);
}
The constructor for symbol_type doesn't take an ::std::string& as
argument, but a constant variant. However, because there is a variant
constructor which takes an ::std::string&, this caused the implicit
construction of a built variant. Considering that the variant argument
for the symbol_type constructor was cv-qualified, this temporary variant
was never destroyed.
As a temporary solution, the symbol was built in two stages:
symbol_type res (token::TOK_TEXT);
res.value.build< ::std::string&> (v);
return res;
However, the solution introduced in this patch contributes to letting
the symbols handle themselves, by supplying them with constructors that
take a non-variant value and build the symbol's own variant with that
value.
* data/variant.hh (b4_symbol_constructor_define_): Use the new
constructors rather than building in a temporary symbol.
(b4_basic_symbol_constructor_declare,
b4_basic_symbol_constructor_define): New macros generating the
constructors.
* data/c++.m4 (basic_symbol): Invoke the macros here.
* data/c++.m4, data/lalr1.cc (by_state, by_type): Do not use -1 to
denote the absence of value, as GCC then fears that this -1 might
be used to dereference arrays (such as yytname).
Use 0, which corresponds to $accept, which is valueless (the needed
property: the symbol destructor must not try to reclaim the memory
associated with the symbol).
* data/c++.m4 (b4_public_types_declare): Declare token_number_type soon.
Introduce symbol_number_type (wider than token_number_type).
Clarify the requirement that kind_type from by_state and by_type
denote the _input_ type (required by the constructor), not the stored type.
Use symbol_number_type and token_number_type where appropriate, instead
of int.
* data/lalr1.cc: Adjust to these changes.
Propagate "symbol_number_type".
Invoke "type_get ()" instead of read "type" directly.
Many 'inline' keywords were in the declarations. They rather belong in
definitions, so move them.
* data/c++.m4 (basic_symbol, by_type): Many inlines here.
* data/lalr1.cc (yytranslate_, yy_destroy_, by_state, yypush_, yypop_): Inline
these as well.
(move): Move the definition outside the struct, where it belongs.
Now that symbols behaves properly, we can eliminate special routines
that are no longer needed.
* data/c++.m4, data/glr.cc, data/lalr1.cc, data/variant.hh:
Remove useless assignment operators and copy constructors.
As a consequence, remove useless includes for "abort".
The current approach was too adhoc: the symbols were not sufficiently
self-contained, in particular wrt memory management. The "new"
guideline is the one that should have been followed from the start:
let the symbols handle themslves, instead of leaving their users to
it. It was justified by the will to avoid gratuitious moves and
copies, but the current approach does not seem to be slower, yet it
will probably be simpler to adjust to support move semantics from
C++11.
The documentation says that the %parse-param are available from the
%destructor. In retrospect, that was a silly design decision, which
we can break for variants, as its a new feature. It should be phased
out for non-variants too.
* data/variant.hh: A variant never knows if it stores something or
not, it is up to its users to store this information.
Yet, in parse.assert mode, make sure the empty/filled variants
are properly used.
(b4_symbol_constructor_define_): Don't call directly the symbol
constructor, to save a useless temporary.
* data/stack.hh (push): Steal the pushed value instead of duplicating
it.
This will simplify the callers of push, who handled this "move"
approach themselves.
* data/c++.m4 (basic_symbol): Let -1, as kind, denote the fact that
a symbol is empty.
This is needed for instance when shifting the lookahead: yyla
is given as argument to "push", and its value is then moved on
the stack. But then yyla must be declared "empty" so that its
destructor won't be called.
(basic_symbol::move): New.
Move the responsibility of calling the destructor from yy_destroy
to ~basic_symbol in the case of variants.
* data/lalr1.cc (stack_symbol_type): Now a derived class from its
previous value, so that we can add a constructor from a symbol_type.
(by_state): State -1 means empty.
(yypush_): Factor, by calling one overload from the other one, and
using the new semantics of stack::push.
No longer reclaim by hand the memory from rhs symbols, since now
that we store objects with proper destructors, they will be reclaimed
automatically.
Conversely, be sure to delete yylhs.
* tests/c++.at (C++ Variant-based Symbols): New "unit" test for
symbols.
* data/c++.m4 (basic_symbol): Keep 'inline' in the prototypes, but don't
duplicate it in the implementation.
* data/variant.hh (variant): 'inline' is not needed when the implementation is
provided in the class definition.
* data/variant.hh (variant, operator=): Make private.
* data/c++.m4 (operator=): New, to avoid needing a definition of that operator
for each class member (such as a possible variant).
* data/glr.cc, data/lalr.cc: Add the necessary include for the abort.
A "symbol" groups together the symbol type (INT, PLUS, etc.), its
possible semantic value, and its optional location. The type is
needed to access the value, as it is stored as a variant/union.
There are two kinds of symbols. "symbol_type" are "external symbols":
they have type, value and location, and are returned by yylex.
"stack_symbol_type" are "internal symbols", they group state number,
value and location, and are stored in the parser stack. The type of
the symbol is computed from the state number.
The class template symbol_base_type<Exact> factors the code common to
stack_symbol_type and symbol_type. It uses the Curiously Recurring
Template pattern so that we can always (static_) downcast to the exact
type. symbol_base_type features value and location, and delegates the
handling of the type to its parameter.
When trying to generalize the support for variant, a significant issue
was revealed: because stack_symbol_type and symbol_type _derive_ from
symbol_base_type, the type/state member is defined _after_ the value
and location. In C++ the order of the definition of the members
defines the order in which they are initialized, things go backward:
the value is initialized _before_ the type. This is wrong, since the
type is needed to access the value.
Therefore, we need another means to factor the common code, one that
ensures the order of the members.
The idea is simple: define two (base) classes that code the symbol
type ("by_type" codes it by its type, and "by_state" by the state
number). Define basic_symbol<Base> as the class template that
provides value and location support. Make it _derive_ from its
parameter, by_type or by_state. Then define stack_symbol_type and
symbol_type as basic_symbol<by_state>, basic_symbol<by_type>. The
name basic_symbol was chosen by similarity with basic_string and
basic_ostream.
* data/c++.m4 (symbol_base_type<Exact>): Remove, replace by...
(basic_symbol<Base>): which derives from its parameter, one of...
(by_state, by_type): which provide means to retrieve the actual type of
symbol.
(symbol_type): Is now basic_symbol<by_type>.
(stack_symbol_type): Is now basic_symbol<by_state>.
* data/lalr1.cc: Many adjustments.
So far we were issuing two lines for each syncline change:
/* Line 356 of yacc.c */
#line 1 "src/parse-gram.y"
This is a lot of clutter, especially when reading diffs, as these
lines change often. Fuse them into a single, shorter, line:
#line 1 "src/parse-gram.y" /* yacc.c:356 */
* data/bison.m4 (b4_syncline): Issue a single line.
Comment improvements.
(b4_sync_start, b4_sync_end): Issue a shorter comment.
* data/c++.m4 (b4_semantic_type_declare): b4_user_code must be
on its own line as it might start with a "#line" directive.
This is mostly used for the license header, the synclines, and the
generated tables:
- /* STOS_[STATE-NUM] -- The (internal number of the) accessing
- symbol of state STATE-NUM. */
+ // STOS_[STATE-NUM] -- The (internal number of the) accessing
+ // symbol of state STATE-NUM.
static const unsigned char yystos_[];
* data/c.m4: Comment changes.
(b4_comment_): Expand the text argument.
Before this change, we were actually formatting M4 code as a
C comment, and then expand it.
(b4_comment): Fix the closing of comments: there is no reason to
add the (line) prefix before the closing "*/".
* data/c++.m4 (b4_comment): New.
The "variant" structure provides a means to store, in a typeless way,
C++ objects. Manipulating it without provide the type of the stored
content is doomed to failure. So provide a means to copy in a type
safe way, and prohibit typeless assignments.
* data/c++.m4 (symbol_type::move): New.
* data/lalr1.cc: Use it.
* data/variant.hh (b4_variant_define): Provide variant::copy.
Let variant::operator= abort.
We cannot undefine it, yet, as it is still uses by the implicit
assigment in symbols, which must also be disabled.
* data/bison.m4 (b4_lex_symbol_if): Rename as...
(b4_token_ctor_if): this.
Depend upon api.token.constructor.
* data/c++.m4, data/lalr1.cc: Adjust.
* doc/bison.texi: Fix all the occurrences of lex_symbol.
* etc/bench.pl.in: Adjust.
* examples/variant.yy: Likewise.
* tests/local.at (AT_BISON_OPTION_PUSHDEFS, AT_BISON_OPTION_POPDEFS):
Handle AT_TOKEN_CTOR_IF.
* tests/c++.at: Adjust to using api.token.constructor and AT_TOKEN_CTOR_IF.
Simplify the test of both build call styles.
(AT_CHECK_VARIANTS): Rename as...
(AT_TEST): this.
And undef when done.
* origin/maint:
tests: check %no-lines
tests: minor simplification
graphs: stylistic changes.
graphs: minor style changes
graphs: show reductions
graphs: style: prefix state number with "state"
graphs: style: use left justification for states
graphs: style: prefix rules and change shapes
obstack: import obstack_finish0 from master
c++: api.location.type
muscles: a function for backward compatibility
maint: more macros
Conflicts:
data/glr.cc
data/java.m4
data/lalr1.cc
doc/bison.texi
src/muscle-tab.c
src/system.h
tests/calc.at
This feature was introduced in 95a2de5695
(which is part of 2.5), but not documented. Give it a proper name, and
make it public.
* data/c++.m4, data/lalr1.cc, data/glr.cc, data/java.m4: Use
api.location.type instead of location_type.
* src/muscle-tab.c (muscle_percent_variable_update): Map the latter to
the former.
* tests/local.at: Adjust.
* tests/calc.at: Use api.location.type.
Leave tests/java.at with location_type, at least for the time being,
to cover both names.
* doc/bison.texi: Document api.location.type.
(User Defined Location Type): New.
* NEWS: Update.
There are currently two systems used to pass information about tokens to
m4: the original one, and another, which is used for instance for
printers and destructors, variants etc. Move to using only the latter.
* data/bison.m4 (b4_symbol_map, b4_token_visible_if)
(b4_token_has_definition, b4_any_token_visible_if, b4_token_format): New.
* data/c++.m4, data/c.m4, data/glr.c, data/java.m4: Adjust to use them.
* origin/maint:
tests: headers.at: strengthen.
glr.cc: do not override C++ definitions by C macros.
YYLLOC_DEFAULT: factor, and don't export it in headers.
api.prefix: do not use #define to handle YYSTYPE_IS_TRIVIAL etc.
tests: portability fixes.
c++: fewer #includes in the headers.
glr.cc: formatting changes.
tests: more logs.
api.prefix: also rename YYDEBUG.
Conflicts:
data/c.m4
data/glr.c
data/glr.cc
data/lalr1.cc
data/yacc.c
* data/c++.m4, data/c.m4 (b4_yylloc_default_define): New.
* data/glr.c, data/glr.cc, data/lalr1.cc, data/yacc.c: Use it.
* data/glr.cc: Do not define YYLLOC_DEFAULT in the header file,
but in the implementation one.
* origin/maint:
bump to 2012 in skeletons.
build: remove ancient Autoconf tests.
doc: c++: complete the location documentation.
c++: locations: provide convenience constructors.
c++: locations: remove useless "inline".
glr: do not use locations when they are not requested
c++: use nullptr for C++11.
build: simplify and improve the compiler warnings for tests.
gnulib: update.
maint: formatting changes.
NEWS: update.
Java: Fix syntax error handling without error token.
tests: beware of -pedantic on large #line numbers.
tests: when using the C++ compiler, use its flags too.
Conflicts:
data/glr.c
data/glr.cc
data/lalr1.cc
data/location.cc
data/yacc.c
tests/Makefile.am
C++11 introduces "nullptr" which plays the role of C's NULL, in
replacement of "0". Fix the C++ skeletons to avoid warnings about
uses of "0" in place of "nullptr", and improve C skeletons to also use
this "nullptr" when compiled with a C++11 compiler.
* configure.ac: More C++ warnings.
* NEWS (2.5.1): Document this.
* data/c++.m4, data/c.m4 (b4_null_define): New.
(b4_null): Use YY_NULL instead of 0.
* data/glr.c, data/lalr1.cc, data/location.cc, data/yacc.c:
Call b4_null_define/b4_null where appropriate.
Use YY_NULL instead of NULL.
* data/location.cc (initialize): Accept a default argument,
YY_NULL.
* tests/actions.at, tests/calc.at: Adjust.
* data/glr.c, lib/libiberty.h, src/system.h (__attribute__):
Do not disable it when __STRICT_ANSI__ is defined, as, for
instance, it disables the __attribute__((unused)) which
protects us from some compiler warnings.
This was already done elsewhere in Bison, in 2001, see
4a0d893695.
* tests/regression.at: Adjust output.
This change was made by applying emacs' untabify function to
nearly all files in Bison's repository. Required tabs in make
files, ChangeLog, regexps, and test code were manually skipped.
Other notable exceptions and changes are listed below.
* bootstrap: Skip because we sync this with gnulib.
* data/m4sugar/foreach.m4
* data/m4sugar/m4sugar.m4: Skip because we sync these with
Autoconf.
* djgpp: Skip because I don't know how to test djgpp properly, and
this code appears to be unmaintained anyway.
* README-hacking (Hacking): Specify that tabs should be avoided
where not required.
* data/bison.m4 (b4_percent_define_use): New.
(b4_percent_define_get): Use it.
Accept a default value.
* data/c++.m4: Do not provide a default value for the %define
variable location_type, rather, use b4_percent_define_get with a
default argument where its value is needed.
* data/lalr1.cc: Do not load location.cc (which outputs both
location.hh and position.hh) if the user defined location_type.
Do not include location.hh either.
* data/glr.cc: Likewise.
(cherry picked from commit 7789b6e3e7)
Conflicts:
ChangeLog
data/bison.m4
data/c++.m4
data/lalr1.cc
* data/bison.m4 (b4_percent_define_get): Accept a default value.
* data/c++.m4: Do not provide a default value for the %define
variable location_type, rather, use b4_percent_define_get with a
default argument where its value is needed.
* data/lalr1.cc: Do not load location.cc (which outputs both
location.hh and position.hh) if the user defined location_type.
Do not include location.hh either.
